(Hypertension. 1998;31:1047-1060.)
© 1998 American Heart Association, Inc.
Review of Alterations in Endothelial Nitric Oxide Production in Diabetes
Protective Role of Arginine on Endothelial Dysfunction
Galen M. Pieper
From the Department of Transplant Surgery, Medical College of Wisconsin,
Froedtert Memorial Hospital (Milwaukee).
Correspondence to Dr Galen M. Pieper, Department of Transplant Surgery, Medical College of Wisconsin, Froedtert Memorial Hospital, 9200 W Wisconsin Ave, Milwaukee, WI 53226. E-mail gmpieper{at}pot.its.mcw.edu
 |
Introduction
|
|---|
Nitric
oxide release from the endothelium plays an important
role in regulation of vascular tone,1 inhibition
of both platelet and leukocyte aggregation and
adhesion,1 2 and inhibition of cell
proliferation.3 These properties suggest that the
level of NO production by the endothelium may
play a pivotal role in the regulation of vascular disease.
Analysis using mass spectrometry has revealed that NO is
produced by NOS from the terminal guanidino nitrogen of the precursor
amino acid L-arginine.4 Thus,
utilization of L-arginine and conversion to NO may
establish a regulatory site in the development of
endothelial dysfunction.
Endothelial dysfunction is characterized by defective
endothelium-dependent relaxation, and some reviews
regarding endothelial dysfunction in diabetes have been
published.5 6 7 8 These reviews have focused on
factors that might contribute to defective relaxation, some of which
will not be addressed in detail in this review. The purpose of the
present review is to summarize evidence that specifically supports
either decreased NO production by diabetic vascular
endothelium and/or impaired NO-mediated
endothelium-dependent relaxation. Second, this review
provides reasonable alternatives to explain some of the controversies
in this research area. Third, since there is growing evidence that
arginine appears to have some benefits for diabetes-associated
abnormalities, this review summarizes the current state of knowledge of
effects of acute and chronic administration of L-arginine
on diabetes-induced endothelial dysfunction and
discusses potential NO-dependent and -independent mechanisms whereby
therapeutic intervention with L-arginine might benefit the
diabetic endothelium.
|
Impaired Endothelium-Dependent Relaxation
|
|---|
Experimental Diabetes Mellitus
Decreases in endothelium-dependent relaxation are
a common feature in both conduit9 10 11 12 13 14 15 16 17 18 19 20 and
resistance21 22 23 24 25 26 27 28 29 30 arteries of chemically induced
experimental diabetic animals (including rats, mice, rabbits, hamsters,
and dogs). In two genetic models of IDDM, similar impaired relaxation
has been documented in aorta31 32 33 and mesenteric
arteries34 35 of diabetes-prone BB/Wor or BB/E
rats and in aorta36 and mesenteric
arteries37 of the diabetes-prone WBN/Kob rat.
Almost without exception, studies that have shown impaired
endothelium-dependent relaxation have found normal
relaxation to nitrovasodilators. These agents relax vascular smooth
muscle by activating guanylate cyclase but, unlike NO, do
not require the presence of the endothelium. Thus, the
intrinsic property to activate vascular smooth muscle
guanylate cyclase appears not to be altered by experimental
diabetes.
Studies in experimental models of NIDDM are few and controversial. In
the obese Zucker rat, no alteration in
endothelium-dependent relaxation has been observed in
intestinal microvessels,38 whereas increased
reactivity has been observed in aorta.39 40 In
contrast, decreased endothelium-dependent relaxation to
acetylcholine but not to A23187 was seen in aorta in male (but not
female) JCR:LA-corpulent rats.41 Little is known
regarding the nature of endothelial dysfunction in
these models.
Conclusion. Collectively, these studies suggest that
endothelial dysfunction (1) is not unique to chemically
induced experimental models of IDDM, (2) is found in both conduit and
resistance arteries, and (3) cannot be explained by an intrinsic change
in reactivity to NO or guanylate cyclase reactivity.
Studies conducted in experimental models of NIDDM are few and provide
no clear consensus regarding this issue.
Human Diabetes Mellitus
The first evidence of endothelial dysfunction in
humans was reported in penile corpura cavernosa of IDDM and NIDDM
patients.42 Impaired
endothelium-dependent relaxation of conduit blood
vessels has been confirmed in both IDDM43 44 45 46 47 and
NIDDM43 48 49 50 51 52 53 patients evaluated in vivo or in
isolated arteries in vitro.45 51 A few studies in
IDDM54 55 56 or NIDDM
patients57 report no defect. One study of
resistance vessels derived from NIDDM patients and evaluated in vitro
showed enhanced reactivity to acetylcholine.58
Evaluation of the latter study is compromised because the control
patients selected exhibited arterial occlusive disease,
hypercholesterolemia, and other nondiabetic
diseases. Also, patients using various cardiovascular
medications were not excluded from the study.
Several investigators have observed normal relaxation to
nitrovasodilators in diabetic
patients.42 43 44 45 46 47 51 52 53 55 56 57 58 59 60 In contrast, other
investigators have noted attenuated responses to nitrovasodilators in
IDDM diabetic patients54 61 62 63 and in NIDDM
diabetic patients.48 49 50 64
Conclusion. The majority of studies have demonstrated
impaired endothelium-dependent relaxation. In contrast,
there is not sufficient consensus on whether vascular smooth muscle
reactivity to NO is compromised in human diabetes mellitus.
|
Importance of Patient Screening and Other Complicating
Factors
|
|---|
One possible explanation of this variability relates to primary
nitrate tolerance, since responses to nitroglycerin but
not sodium nitroprusside were diminished in diabetic
patients.65 Alternatively, one of these studies
noted no difference in reactivity to nitroglycerin
after correction for differences in baseline
diameter.61 Furthermore, diabetic patients with
microalbuminuria exhibited impaired dilation to isosorbide
dinitrate, while in diabetic patients without
microalbuminuria, dilation was
unaltered.66 In this regard, dilation in response
to carbachol was normal in diabetic patients but was blunted by an NOS
inhibitor in both control and diabetic patients without
microalbuminuria but not in diabetic patients with
microalbuminuria.56 Thus, there may
be variable mechanisms of impaired vascular reactivity to
nitrovasodilators among different diabetic patient subpopulations.
There are several concerns regarding critical evaluation of the data
obtained in diabetic patients. Whereas some investigators have
carefully eliminated patients taking medications (eg, vasoactive and
cardiovascular drugs) and have limited intake of
caffeine or alcohol before evaluation, other investigators have either
not provided information regarding patient selection or have ignored
these potential influences. Some studies have included patients with
histories of smoking (an independent risk factor for
endothelial dysfunction) or have only limited smoking
just before the study.
Most studies have used a mixed-gender patient population. Thus, the
influence of gender-related differences has not been isolated. Indeed,
it is widely known that diabetes-induced complications are enhanced in
the female population. A recent study has revealed diminished reactive
hyperemia (an endothelium-related phenomenon)
in young female IDDM patients compared with age-matched male
subjects.67 Furthermore, inclusion of postpuberty
female patients using oral contraceptive agents has been acknowledged
in some studies,61 62 but the significance of
inclusion of this subpopulation has not been addressed. In studies
using male subjects only, endothelial function was
either impaired46 47 49 50 or
unchanged.54 55 57
In studies of vascular reactivity using NIDDM patients, it was
acknowledged that many subjects were receiving hypoglycemic drugs or
insulin at the time of
evaluation.47 48 49 51 52 64 68 Oral hypoglycemic
agents such as glibenclamide can impair dilation mediated by
NO-mediated KATP channel
activation.69 Thus, the effects of this
medication on NO-mediated relaxation have not been uniformly excluded.
One study showed no difference in endothelium-dependent
relaxation between patients taking oral hypoglycemic agents and those
controlling the disease by diet alone.48 In one
study in which hypoglycemic medication was removed for a finite period
before evaluation, there was diminished reactivity to acetylcholine but
not to nitroprusside.53
Endothelial function in patients has been determined in
vivo using either strain-gauge plethysmography or Doppler
techniques, predominately in brachial/forearm protocols. A few have
evaluated dilation in the femoral artery53 63 64
or coronary arteries.43 While there is
important value in examining reactivity in vivo under ambient
conditions, there are several significant complicating factors that
make the in vivo delineation of mechanisms difficult. For example, one
investigator found that endothelial dysfunction
correlated with serum triglyceride and lower HDL
cholesterol levels.50 Furthermore,
ambient glycated proteins may diminish
endothelium-dependent relaxation studied in
vitro,70 71 although a recent in vitro study in
canine conduit and resistance artery preparations could not confirm
this effect.72
Conclusion. Patient selection and screening have the
potential to explain differences in findings obtained in clinical
studies. The potential influence of gender, presence of
microalbuminuria, and use of medications, including oral
contraceptives and hypoglycemic agents, need to be rigorously evaluated
in clinical studies. Also, the influence of ambient glucose
concentration, glycated proteins, and lipids on
endothelial function needs to be considered in clinical
evaluations.
|
Confounding Effects of Ambient Glucose Concentration
|
|---|
Ambient plasma glucose concentration could be a confounding factor
in vivo as well. Some studies in experimental animal models have noted
acute vasodilation (ie, seconds to minutes) in response to concurrent
elevation in glucose concentration.73 74 75 In
contrast, short-term exposure (ie, minutes to several hours) to
elevated glucose concentrations in vitro or topical application in situ
impairs basal NO tone76 and agonist-stimulated
endothelium-dependent
relaxation.77 78 79 80 81 Interestingly, 24-hour infusion
of glucose in normal subjects did not impair
relaxation.82 Because multiple agonists were used
in the latter study, the significance of masking by crossover effects
of individual drugs cannot be excluded in the interpretation of this
data. In contrast, evaluation during an acute bolus of glucose
administration (ie, during a glucose tolerance test) revealed decreased
endothelium-dependent
relaxation.83
An immediate acute effect of elevated glucose concentration is a marked
increase in basal [Ca2+]i
and NO production in isolated human endothelial
cells.84
[Ca2+]i also increases
rapidly (within minutes) in the presence of elevated glucose
concentration in isolated bovine aortic endothelial
cells (G.M.P., unpublished observation, 1997). This effect may persist
for several hours after exposure to elevated glucose
concentration.85 In contrast, after 24 hours of
exposure of bovine aortic endothelial cells to elevated
glucose and subsequent analysis under normal glucose
conditions, basal [Ca2+]i
and NO production returned to normal, whereas
bradykinin-stimulated
[Ca2+]i and NO
production was reduced.86 Similarly,
impaired agonist-stimulated
[Ca2+]i and NO
production have been observed after exposure of rat or
porcine endothelial cells to elevated glucose
concentration.87 88
To circumvent this potential limitation, one study using the
euglycemic insulin clamp protocol46
revealed that endothelial function was diminished.
Also, analysis in vitro of human gluteal resistance arteries
under normoglycemic conditions also revealed diminished
endothelium-dependent
relaxation.45 The latter study suggested an
intrinsic defect in endothelium-dependent relaxation in
IDDM that is independent of direct effects of concurrent ambient
glucose concentrations in vivo. In contrast, another study showed that
relaxation to carbachol was unchanged in IDDM patients under
euglycemic conditions produced by insulin
infusions.56 Interestingly, a defect in NO
synthesis was still suggested in patients with
microalbuminuria based on the lack of response to
NG-monomethyl-L-arginine
(L-NMMA).
Conclusion. The confounding effects of ambient glucose
concentration on endothelial function in studies
performed in patients and in experimental diabetes in vivo should be
taken into account when glucose concentration is not normalized.
Unfortunately, controlling for this contingency may be difficult,
particularly in clinical studies. Clearly, there are important direct
stimulatory and inhibitory effects of elevated glucose
concentration on vascular reactivity that may be manifested at
different times subsequent to rises in glucose concentration.
|
Paradoxical Findings on Endothelial Function: Role
of Disease Duration
|
|---|
Few studies have examined the temporal nature of the onset of
endothelial dysfunction in experimental diabetes. Most
have shown a progressive worsening of dysfunction that appears to
plateau at some finite point in time. Dysfunction has been reported at
as early as 1 week of diabetes in rat intestinal
arterioles,30 after 2 weeks in hindquarters but
not in mesenteric or renal arteries,89 after 3
weeks in cremaster muscle arterioles,90 after 4
to 6 weeks in mesenteric arterioles,25 91 and
after 4 weeks in aorta.70 Thus, lack of
endothelial dysfunction of diabetic rat aorta at 2 to 3
weeks of disease92 could be explained by the
short period of time studied. It is important to note that the onset of
endothelial dysfunction may vary widely among
individual vascular beds and/or the severity of the diabetic model used
in any given study.
Interestingly, there are paradoxical reports of normal
endothelium-dependent relaxation in aorta after >12
weeks of diabetes93 94 95 and after 15 to 17 weeks
of diabetes in perfused mesenteric
preparations.96 The reasons for these disparate
observations compared with other findings are unclear. Potential
explanations might include the use of helical
strips,93 diabetes-induced hypersensitivity to
phenylephrine and evaluation only in
indomethacin-treated
preparations,94 use of a low dose of
streptozotocin,94 or the potential influence of
masking by crossover effects due to multiple drug
challenges.95
There is clinical and experimental evidence showing augmented blood
flow at early stages of diabetes.97 It is not at
all clear whether this increased blood flow reflects increased
NO-specific endothelium-dependent dilation. One
possibility that needs to be examined is whether the diabetes-induced
decreases in 2,3-diphosphoglyceric acid levels in red blood
cells,98 which regulate oxygen release from
hemoglobin, lead to a "hypoxic-like"
environment.99 Decreases in tissue ATP
concentration could result in increased blood flow due to
hypoxic-induced dilation in compensation for reduced oxygen delivery to
tissue. Indeed, myocardial ATP concentration is reduced in diabetes but
is rapidly replenished within minutes of perfusion in vitro with
oxygenated blood-free salt
solutions.100 Furthermore, in situ
analysis using microelectrodes reveals diminished oxygen
tension in aorta of alloxan-diabetic rabbits.101
Thus, increase in tissue blood flow at the early stages of diabetes may
be due to hypoxic vasodilation. Alternatively, increased blood flow may
be a direct response to acute or short-term hyperglycemia, since
coronary blood flow is increased with glucose infusions in
isolated hearts in the presence of
indomethacin.75
At least one study has noted augmented
endothelium-dependent relaxation in
indomethacin-treated rat renal arteries at an early
stage of diabetes.102 Because
endothelium-dependent relaxation is reduced in renal
arteries of diabetic rats of longer disease
duration,103 this observation raises the
possibility that diabetes induces biphasic effects on
endothelium-dependent relaxation. Thus, an early
increase in blood flow may be followed by a transition state to
impaired relaxation. In support of this hypothesis, one study noted an
increase in endothelium-dependent relaxation of
mesenteric arteries to both acetylcholine and bradykinin at 6 weeks of
diabetes that reverts back to normal after 12
weeks.104 This enhanced relaxation could not be
accounted for by intrinsic changes in smooth muscle reactivity because
responses to sodium nitroprusside were normal.
It remains to be resolved in temporal studies using various
preparations whether the enhanced endothelium-dependent
relaxation at early stages might be due to diabetes-induced increases
in synthesis of vasodilator prostaglandins, EDHF, or NO, or
a combination of any of these endothelium-derived
factors. In this regard, studies conducted in
mesenteric104 and renal102
arteries in the presence of indomethacin suggest that
vasoactive prostanoids do not contribute to enhanced
endothelium-dependent relaxation, leaving EDHF and NO
as candidate factors.
Conclusion. Time of evaluation after onset of disease
may be critical to demonstrating endothelial
dysfunction. Indeed, evidence exists for enhanced
endothelial function at early stages followed by
dysfunction at later stages. Whether these opposing actions are
causally linked is not yet known. Further studies are warranted to
understand this temporal dichotomy.
|
Role of Prostanoids in Endothelium-Dependent
Dilation in Diabetes
|
|---|
Impaired endothelium-dependent relaxation in
diabetes cannot always be assumed to be mediated by a reduction in NO
activity or synthesis, since some vasodilators also release
prostaglandins. Changes in prostaglandin
synthesis may alter NO production or reactivity to NO. Because
diabetes alters the reactivity to prostanoids and may either increase
or decrease prostacyclin production depending on the artery
chosen,105 this issue needs to be resolved,
especially at various stages of disease and in individual blood vessel
preparations.
Simultaneous enhancement and release of vasoconstrictor
prostaglandins may explain some instances of impaired
endothelium-dependent relaxation. Indeed, in aorta,
pial artery, and mesenteric artery, relaxation is either
normal94 or is normalized in the presence of
indomethacin or thromboxane receptor
antagonists.13 21 103 106 In
contrast, endothelial function was normal in
coronary arteries of alloxan-diabetic dogs, but dysfunction was
unmasked in the presence of inhibitors of
cyclooxygenase,11 107
suggesting enhanced compensatory increases in vasodilator prostanoid
release. Furthermore, increases in production of the
vasodilator prostacyclin have been reported in perfused mesenteric beds
of 3-week diabetic rats.108 While not directly
examined, this might provide an alternative explanation for one of the
early reports showing augmented relaxation to the agonist histamine in
mesenteric arteries of diabetic rats109 or the
observation of normal relaxation seen in perfused mesenteric beds of
long-term diabetic rats.94 In contrast, this may
not be adequate to explain the increased
endothelium-dependent relaxation seen in perfused
kidney at early stages of disease, since this increase was seen in
indomethacin-treated
preparations.102
One investigator has shown that thromboxane receptor
antagonism does not alter relaxation in basilar
artery110 but restores relaxation of pial
arteries taken from the streptozotocin-diabetic rat
model.21 These observations suggest important
regional differences, although one cannot exclude the possible
contribution of the superimposition of the variable in the duration
of disease, which was 4 to 5 months in the basilar artery study versus
2.5 to 3.5 months for the pial artery study. In another model of
alloxan-diabetic dogs in which duration of disease was held constant,
inclusion of indomethacin or ibuprofen unmasked a
defect of endothelium-dependent relaxation in
coronary artery11 but not renal
artery.111 Collectively, these studies suggest
important regional differences in the mechanism of
endothelial dysfunction.
Many studies using either rat conduit
arteries9 14 16 17 18 112 113 114 or rat resistance
arteries25 26 91 103 115 116 117 have indicated no
improvement in endothelial function after evaluation
under conditions of cyclooxygenase blockade or
thromboxane receptor antagonism. Similar interventions have
failed to modify defective relaxation in aorta of the genetic diabetic
BB rat33 and in coronary arteries of
alloxan-diabetic dogs.20
The effect of prostanoids on endothelial function in
human diabetes has not been routinely evaluated. Prior treatment with
indomethacin in vitro normalized impaired
endothelium-dependent relaxation in cutaneous arteries
of gestational diabetes.118 In contrast, in one
human study in which all patients received aspirin before
evaluation,44 the authors concluded that
exogenous prostanoid synthesis cannot account for the
endothelial dysfunction in IDDM.
Conclusion. There exist some potentially important
regional differences in the role of prostanoids in contributing to
altered endothelium-dependent dilation in diabetes
mellitus, although this cannot be easily predicted among various
conduit versus resistance vessels. Nevertheless, there is clear and
ample evidence to suggest that alterations in prostanoid
production may not always account for and/or may not be
obligatory for impaired endothelial function in
diabetes.
|
NO-Dependent or -Independent Endothelial Dysfunction
|
|---|
Implicit in all of these studies is the assumption that
endothelium-dependent relaxation in both control and
diabetic blood vessels is exclusively mediated via NO. This assumption
is hazardous because endothelium-dependent relaxation
to certain agonists and in certain arteries appears to be mediated in
part by vasodilator prostanoids or by an EDHF.119
The entity of EDHF is not known with complete certainty, but at least
one EDHF is believed to be an epoxide of arachidonic
acid that is formed by a cytochrome P450–derived
monooxygenase.120
EDHF appears to activate K+ channels,
especially calcium-activated K+
channels.119 The contribution of
KATP channels to relaxation in diabetes is
uncertain and should also be considered. A few studies have noted
diminished responses to KATP channel openers in
diabetic rat aorta15 121 122 and basilar
artery.123 One study in the aorta of WBN/Kob rat
showed no alteration in relaxation to the KATP
channel agonist cromakalim.124 In contrast,
others have observed a paradoxically enhanced response in dog
coronaries to aprikalim, albeit in short-term
diabetes.125 In normal arteries, it is generally
believed that the component of EDHF versus NO that contributes to total
relaxation increases with decreasing vessel size. The observation that
diabetes-induced endothelial dysfunction can occur in
both conduit and resistance arteries despite
cyclooxygenase blockade suggests that this
dysfunction could be explained by defects in EDHF or by deficits in NO
synthesis unrelated to or in addition to changes in EDHF or
prostanoids.
Two studies in rat aorta reveal that endothelial
dysfunction persists despite pretreatment with TEA (to inhibit
calcium-activated K+ channels),
suggesting that defects in EDHF may not be
operative.126 127 Alternatively, a recent study
in perfused kidney indicates that endothelium-dependent
relaxation in control kidney arises from both NO and EDHF, whereas
relaxation in diabetic kidney arises from NO, EDHF, and
prostanoids.128
Few studies have examined perturbations in membrane polarization in
diabetic vascular tissue. Membrane
hyperpolarization is known to occur in gastric
gland of 2- to 3-day diabetic rabbits129 and in
endothelial cells from human subjects with gestational
diabetes.130 One study showed normal resting
membrane potential but diminished hyperpolarization
in the response of diabetic mesenteric artery to
acetylcholine131 despite unaltered
hyperpolarization in response to the
K+ channel agonist pinacidil. This report
conducted in the presence of NOS and cyclooxygenase
inhibitors suggested diminished
endothelium-dependent, TEA-sensitive
hyperpolarization and relaxation.
Conclusion. Endothelial dysfunction in
some cases of diabetes and in certain blood vessel types may arise from
deficits in EDHF, but it is also clear that endothelial
dysfunction can also occur despite blockade of EDHF and prostanoid
synthesis and action, suggesting a role for deficits of
endothelium-derived NO. It is possible that previous
studies may need to be carefully reevaluated, since alternative
compensatory pathways including cytochrome P450–derived EDHF may be
activated or inactivated, which could mask impaired
endothelium-dependent relaxation or compensate for
defective NO synthesis. It should be emphasized that parallel
compensatory pathways may be important at certain stages but perhaps
not at all stages of the disease.
|
Evidence Supporting Altered NO Production From
Diabetic Endothelium
|
|---|
Functional Studies in the Presence of SOD
NO activity is known to be reduced by chemical interaction and
destruction by superoxide anion radicals. Several investigators have
noted improved relaxation after acute incubation with
SOD.14 25 26 33 77 117 132 This issue has been
discussed in detail in a previous review.7
Nevertheless, these studies are consistent with increased
destruction of NO and decreased bioactivity of NO in diabetes, but they
do not give adequate direct information regarding perturbations in NO
production.
Currently, there have been no studies that have directly determined NO
synthesis or release from diabetic endothelium. Only
one laboratory has examined intraluminal release of NO activity from
perfused diabetic rat aorta donor segments using the bioassay
technique.133 Accordingly, luminally released
basal NO activity in perfused diabetic rat aorta was normal, but
addition of SOD caused a larger incremental increase in relaxation of
the bioassay detector when the diabetic donor segment was used. This
suggested the increased release of superoxide anion radicals from
diabetic rat endothelium. In contrast,
acetylcholine-stimulated endothelium-derived relaxing
factor/NO bioactivity from perfused diabetic rat aorta was diminished
but also normalized by perfusion with SOD.114
Conclusion. Several studies support the concept that
IDDM decreases NO bioactivity. This may be due in part to enhanced
destruction of NO by increased superoxide synthesis.
Indirect Evidence Using Guanylate Cyclase
Inhibitors
Use of inhibitors of NOS or NO reactivity (eg,
hemoglobin or methylene blue) has given indirect information regarding
the contribution of NO to basal tone and agonist-induced relaxation in
control versus diseased blood vessels. Many studies that have used
removal of endothelium or NOS inhibitors or
hemoglobin, which increase the precontracted tone of arteries, reveal
decreased basal NO activity in both diabetic
conduit16 94 134 135 136 137 138 139 140 and
resistance24 25 35 132 140 141 142 143 144 145 arteries under
both in vitro and in vivo conditions.
Two studies noted enhanced contraction to methylene blue in control
versus diabetic rat aorta ring preparations, suggesting diminished
basal NO in diabetic arteries.16 133 The use of
methylene blue alone is inadequate because of the known effects of
methylene blue to stimulate superoxide anion radical or inhibit
NOS.146 147 One study using a new highly
selective inhibitor of guanylate cyclase, ODQ,
revealed that ODQ completely inhibits acetylcholine-induced relaxation
in control rat aorta and inhibits relaxation in diabetic rat aorta by
80%.148 This suggests that almost all of the
relaxation in both groups is mediated by cGMP activation, implicating
NO as the EDRF. Also, the small ODQ-resistant component of
relaxation in diabetic rat aorta leaves open the possibility of some
other unknown factor that appears unrelated to vasodilator prostanoids,
H2O2, or a TEA-sensitive
EDHF.
Conclusion. Overall, these studies suggest diminished
basal NO production in diabetes.
Evidence Using NO Trapping Agents
Basal NO activity in rat aorta has also been assessed using either
a small-molecular-weight, iron-thiol-based NO
scavenger149 or a nitronyl nitroxide trapping
agent.148 These agents are different in that they
do not alter NOS activity but react with NO after its release. These
probes were effective in scavenging all basal NO based on similar
tension responses after removal of endothelium or by
using NOS inhibitors and confirm a smaller basal NO
activity in diabetic rat aorta. This NO scavenger–sensitive basal NO
activity was augmented by L-arginine in diabetic but not
control aorta.149
In contrast, despite nearly complete inhibition of
acetylcholine-mediated relaxation by both control and diabetic rat
aorta by NOS inhibitors, use of a nitronyl nitroxide
revealed an NO scavenger–resistant component of
agonist-stimulated relaxation that is greater in diabetic than control
arteries.148 This resistant component of
endothelium-dependent relaxation was sensitive to NOS
inhibitor but was insensitive to
indomethacin, TEA, or catalase, suggesting that this
additional EDRF is likely not EDHF, prostanoids, or
H2O2. A significant portion
of this resistant component was eliminated using ODQ,
suggesting that this EDRF is an activator of
guanylate cyclase. Current research continues to attempt to
identify whether diabetic arteries produce an additional vasoactive
product that is derived from the NOS reaction but that may not
necessarily be a free NO radical.
Measurement of NO by cGMP Generation
Information regarding potential deficits in NO synthesis have also
been derived from measurement of cGMP in vascular tissue. Although one
study showed no significant alteration in basal and
acetylcholine-stimulated cGMP in alloxan-diabetic
aorta,150 an earlier study in the same animal
model showed that diabetes decreased both basal and
acetylcholine-stimulated cGMP.19 The latter study
differed only in that duration of disease was longer and that a
preconstrictor agonist was used in the assays. Similar decreases in
acetylcholine-stimulated cGMP production have been observed in
rat and rabbit aorta and mesenteric
arteries,12 19 151 152 in canine coronary
arteries,107 and in isolated rat
glomeruli.153 154 In contrast, cGMP in response
to nitroprusside was diminished in isolated glomeruli from diabetic
rats.153 In the presence of the phosphodiesterase
inhibitor IBMX and after stimulation with nitroprusside,
cGMP production was not altered by diabetes in rat
aorta152 or
glomeruli.154
There is no apparent intrinsic change in either guanylate
cyclase or phosphodiesterase activity of vascular smooth muscle to
account for defective cGMP production in diabetic blood
vessels. The evidence suggests decreased basal and agonist-stimulated
NO bioactivity in experimental diabetes mellitus.
Measurement of NO or NO By-products
To our knowledge, there has been no direct measurement of NO from
diabetic arteries using NO electrodes or an NO analyzer.
According to indirect measurements of the NO-based by-products
nitrate and nitrite, levels of either nitrite or nitrate+nitrite
in urine of diabetic rats were
increased.26 142 155 156 157 158 159 160 161
Streptozotocin, the agent frequently used to produce experimental
diabetes, is also a known NO donor compound.162
This property could potentially contribute to enhanced urinary
nitrate/nitrite levels at early stages of diabetes. This limitation is
likely not a problem for samples taken several weeks after the
administration of streptozotocin. Furthermore, the increased urinary
nitrate+nitrite might simply be secondary to hyperfiltration. Although
some studies have controlled for nitrate/nitrite derived from food
during diabetes-induced hyperphagia, most have not controlled for
variation in nitrate consumption derived from drinking water due to
diabetes-induced increases in water consumption or for NO released
simply because of the mechanical stimuli of increased renal blood flow
and filtration due to excess volume disposal as a consequence of
diabetes. Thus, the relationship between urinary nitrate/nitrite
concentration and vascular endothelium
production of NO must be understood in the context of these
significant limitations.
Some studies have reported increases in both urine and plasma
nitrate+nitrite concentration in diabetic
animals.157 158 159 161 In contrast, levels of
plasma nitrate163 and of
cGMP164 were normal in diabetic patients.
Although nitrate levels were normal,163 exhaled
NO level was decreased by 30% but did not reach statistical
significance because of the small sample size. Infusion of L-NMMA
caused a greater decrease in exhaled NO in healthy versus diabetic
patients, although nitrate levels did not change, suggesting decreased
NO synthesis. Nevertheless, changes in plasma NO do not provide
concrete information regarding the location and source of NO
production.
Conclusion. Measurement of nitrate/nitrite in plasma and
urine in experimental diabetes supports increased NO production.
Because the source of this increased nitrate and nitrite is not easily
ascertained, conclusions regarding endothelial cell NO
production based on these measures alone are hazardous and must
be taken as confirmation of other derived parameters.
Measurement of NOS Protein, mRNA, or Activity
There is little information available concerning measurement of
endothelial cell NOS protein content or mRNA in
diabetes mellitus. Diaphorase staining and
immunohistochemical staining suggest reduced NOS activity in
sympathetic autonomic ganglia165 and macula
densa166 of experimental diabetic rats. Another
study revealed a decrease in mRNA and content of neuronal NOS (nNOS) in
diabetic penile tissue.167 Interestingly, one
study in diabetic rat heart homogenates noted increased NOS
activity (believed to be largely constitutive NOS [cNOS]) determined
by arginine to citrulline conversion along with increased cNOS mRNA and
protein.26 There was no detectable inducible NOS
(iNOS), although induction of iNOS was previously reported in placenta
of gestational diabetic patients.168 In rat renal
tissue at 1 week of diabetes, mRNA and protein for cNOS, iNOS, and nNOS
were all increased in cortex but unchanged in
medulla.169
Consistent with these reports are increases in mRNA and/or
protein for cNOS after culture with elevated glucose concentrations in
bovine aortic endothelial cells for 24 hours (G.M.P.,
unpublished observations, 1997) and in human aortic
endothelial cells after 5
days.170 In contrast, longer-term culture under
high glucose conditions of human umbilical vein
endothelial cells revealed no alteration in either mRNA
or protein for cNOS.171 This suggests either
heterogeneity in the response of different
endothelial cells or glucose concentration–dependent
and/or exposure time–dependent differences.
In the diabetic heart study, the finding that normal or enhanced NOS
activity occurs under acellular conditions despite concomitant
decreases in endothelium-dependent relaxation indicates
no intrinsic defect in NOS activity. This is consistent with
elevations in mRNA for endothelial cell–derived NOS despite decreased
NO production in endothelial cells cultured
with elevated glucose for 24 hours.86
In isolated endothelial cells from diabetic BB
rat,172 nitrate/nitrite production in
culture was diminished, as well as citrulline production from
arginine, and occurred without changes in arginine uptake. The latter
suggests no defect in the y+ transporter function
in these cells. In contrast, alloxan diabetes enhanced cGMP
production and increased both Vmax and
Km for arginine in gastric
glands.129 The significance of these observations
to arginine content in vascular disease is unknown; however, it should
be noted that these studies were conducted just 2 to 3 days after
diabetes induction. Similar actions have been found by this same group
in human fetal endothelial cells in gestational
diabetes.130
Conclusion. There are too few measurements of NOS
activity, mRNA, and protein in the literature to formulate a consensus.
There is some evidence of increased mRNA and protein despite impaired
endothelial function. These observations suggest that
NO synthesis might be regulated by factors present in the intact
cell under conditions (eg, substrate provision or cofactor
concentration) that may not be optimal for NOS enzyme activity (see
"Discussion").
|
Arginine Deficiency and/or Substrate Utilization by NOS
|
|---|
Plasma arginine concentration is decreased in diabetic
rats26 138 159 173 174 and in some human diabetic
patients.175 176 In contrast, some reports
involving a small number of patients have reported normal arginine
levels.177 178 In streptozotocin-induced
diabetes, decreases in plasma arginine and endothelial
dysfunction are corrected by subsequent transplantation of either whole
pancreas179 or pancreatic
islets.174
The reasons for decreased plasma arginine are unclear. One study showed
that plasma arginine was decreased at 14 days.159
This was associated with a larger excretion of arginine with increases
in nitrate/nitrite and cGMP, suggesting that lower plasma arginine
results from increased renal NO production and/or increased
elimination of arginine. Tissue levels of arginine might be decreased
as the plasma arginine concentration in diabetic rats is near the
Km for arginine transport into
endothelial cells and the
Km does not appear to change, at least in
certain tissue and during early stages of
diabetes.129
There is a paucity of information regarding arginine content in
diabetic tissue. Despite increased arginine content in gastric glands
of 2- to 3-day alloxan-diabetic rabbits,20 the
content of arginine in freshly isolated diabetic rat aorta is
reduced.127 If arginine content is in fact
diminished in various vascular tissue in diabetes, then replacement
therapy should improve NO-mediated relaxation. One study showed that
L-arginine partially improved, albeit modestly, basal and
agonist-stimulated cGMP in diabetic glomeruli.154
Incubation in vitro with L-arginine but not
D-arginine augments relaxation to acetylcholine but not to
nitroglycerin in both
streptozotocin138 174 and BB
diabetic180 rat aorta. The fact that
L-arginine supplementation increased NOS
inhibitor–sensitive, acetylcholine-stimulated cGMP
production, but not nitroglycerin-stimulated
cGMP production, indicates that this effect is specific for
augmenting endogenous NO
synthesis.127 In contrast, in one study
L-arginine did not improve relaxation to acetylcholine in
diabetic rat aorta.180 It is difficult to
reconcile these opposite findings because the duration of disease,
species, and artery type were similar. Differences in experimental
design and protocols, arginine concentration, and duration of arginine
exposure may explain these varied findings. For example, in the study
in which arginine had no effect,181 arginine was
given between the second and third challenges to acetylcholine. It
would be interesting to know whether multiple challenges in vitro in
arginine-free buffers with agents that stimulate NO release further
deplete the reservoir of arginine in diabetic tissue such that reversal
by L-arginine becomes more resistant.
It has been suggested99 182 that the beneficial
effects of L-arginine are specific to isolated preparations
and due to a theoretical depletion of tissue arginine, since arginine
is not normally present in the buffers used for in vitro vascular
function studies. This hypothesis is based on an earlier
study183 and is confirmed
elsewhere,184 wherein arginine content decreases
rather rapidly in endothelial cells cultured under
arginine-free conditions. This possibility cannot yet be excluded for
the various in vitro diabetic vascular studies because this has not yet
been tested. This hypothesis has been challenged8
because dysfunction occurs in animals and humans at
physiological arginine concentrations.
There are several points that mitigate against the in vitro hypothesis.
First, while incubation of porcine aortic endothelial
cells in arginine-free media results in depletion of arginine content,
this phenomenon apparently does not occur in all types of
endothelial cells, even for durations of up to 24 hours
of arginine-free conditions.185 Therefore, it
cannot be automatically assumed that arginine depletion in
endothelial cells occurs in all intact artery
preparations in vitro. Second, if the endothelial
dysfunction seen in diabetic arteries tested in vitro is merely due to
the absence of arginine in the buffer, it is difficult to reconcile why
one does not always observe diminished relaxation in all studies and to
all endothelium-dependent vasodilators that also act
via NO. Indeed, in the absence of arginine-supplemented buffer,
receptor-independent, endothelium-dependent relaxation
to the calcium ionophore A23187 in aorta that is normally blocked fully
by NOS inhibitors is not diminished in either
streptozotocin-induced137 or genetically prone BB
diabetic rats.33 Third, tissue arginine content
in diabetic rat aorta is decreased despite the fact that samples were
freshly extracted from animals (ie, without extended incubation in
arginine-free media). Fourth, arginine content in isolated
coronary endothelial cells of the diabetic BB
rat was normal or even slightly elevated, yet these same cells
demonstrated defective NO synthesis based on both nitrite release into
media and arginine-to-citrulline conversion
analysis.172 These studies also showed
that NOS activity in cell homogenates was normal,
suggesting impaired arginine utilization under intact cell conditions.
Fifth, acetylcholine-stimulated relaxation of diabetic canine
coronary arteries was decreased in vivo but not in
vitro,107 and addition of L-arginine
to diabetic canine coronary arteries in vitro did not enhance
the diminished acetylcholine-stimulated cGMP
production.107 In contrast, in other
studies, intracoronary infusion of L-arginine
infusion in diabetic dog hearts improved
endothelium-dependent
relaxation,20 and in situ application of
L-arginine improved nerve blood flow in the
rat.186
These observations would argue against effects of
L-arginine that are unique to the in vitro studies and
unique to the rat model. Interestingly, effects in the diabetic dog in
vivo occurred despite normal plasma arginine concentrations, suggesting
that static plasma arginine levels need not be reduced for there to be
improvement in endothelial function by
L-arginine administration. Why this occurs is unknown. One
possibility is that arginine transport might be reduced directly by
elevated glycated proteins in the blood.187
There are reports that appear not to support the concept of defects in
L-arginine utilization. In one case, topical application of
L-arginine apparently did not improve relaxation in basilar
artery of streptozotocin-induced diabetic
rats,188 in cheek pouch microcirculation of
diabetic hamsters,189 or in mesenteric artery of
BB diabetic rats.34 Additional information is
needed to determine with total certainty whether these studies reflect
important regional differences in the beneficial action of
L-arginine on reversal of diabetic
endothelial dysfunction because of differences in
arginine concentration, duration of exposure, location of application,
and duration of disease (see "Discussion") and the potential for
crossover effects due to multiple agonist challenges in the same
preparation in at least two of these
studies.34 189 In addition, the ineffectiveness
of L-arginine in mesenteric arteries of BB rats is
inconclusive, since a short incubation period was used as well as a
concentration (ie, 10 µmol/L) that is 20 times lower than the
physiological concentration of arginine in rat
plasma and 10 times lower than the Km of
arginine transport into endothelial
cells.129 Furthermore, the potential of masking
by crossover effects due to multiple agonist administration has not
been excluded in these studies.
One should not discount the possibility that progressive changes in
pathology may make endothelial dysfunction
resistant to modification by L-arginine
administration. Indeed, arginine has been shown to be beneficial in
short-term190 but not long-term
atherosclerosis.191 Indeed,
L-arginine administration in vitro restores
endothelium-dependent relaxation after 8 weeks of
diabetes but not after 12 weeks of diabetes in aorta of Lewis-strain
diabetic rats.174 It may be noteworthy to
consider the possibility that the negative results of topical
application of L-arginine in situ in basilar arteries were
observed in animals of 12-week disease duration. Currently, there is no
information whether duration of disease alters arginine transport
kinetics.
Transport kinetics into vascular smooth muscle cells during topical
application of arginine in diabetic arteries is another consideration.
This initial uptake must be sufficient to optimize intracellular
arginine at a concentration adequate to facilitate subsequent uptake by
diabetic endothelium. It was presumed that the
concentration of arginine that failed to augment relaxation in diabetic
rat basilar artery and hamster cheek pouch microcirculation was
adequate on the basis of improved relaxation in basilar artery of
hypertensive rats192 and in cheek pouch of
cardiomyopathic hamsters193 using
a similar arginine concentration. It is possible that the concentration
and time of exposure and alterations in transport into vascular smooth
muscle are complications that need to be considered for diabetic
tissue. Indeed, in studies conducted within the same
species,194 the Km
for transport into smooth muscle cells is much lower (ie, 25 µmol/L)
compared with that for endothelial cells (ie, 
100
µmol/L). Furthermore, the maximum capacity for arginine transport is
much lower for smooth muscle compared with endothelial
cells. These effects might limit the amount of arginine transported by
smooth muscle cells and subsequently delivered to
endothelial cells to improve NO synthesis in topical
application protocols.
Alternatively, protein kinase C (PKC) is known to be increased in
diabetic rat aorta and heart195 196 and in
vascular smooth muscle cells exposed to elevated glucose
concentrations.196 Furthermore, PKC
inhibitors normalize relaxation in the cerebral arteries of
diabetic or hyperglycemic rats.80 197 PKC
activation is known to markedly reduce arginine transport in normal rat
vascular smooth muscle cells.198 Thus, these
alternative hypotheses should be addressed before final conclusions can
be drawn regarding regional differences in efficacy of arginine to
restore endothelial function.
The efficacy of arginine in diabetic kidney seems less in doubt. The
hyperfiltration and proteinuria produced in short-term diabetes in rats
was prevented by dietary arginine
supplementation.158 Paradoxically, this was
associated with a reversal of the enhanced urinary excretion of
nitrate/nitrite and cGMP, suggesting that dietary arginine represses
the augmented renal production of NO at early stages of the
disease. In 2-month diabetic rats, arginine modestly improved the
decreased basal and agonist-stimulated cGMP in glomeruli but not to
control levels (cGMP). Interestingly, with slightly longer disease
duration, arginine failed to alter the decreased basal and
agonist-stimulated NO as measured by the NO
electrode,199 which is in agreement with temporal
studies in diabetic rat aorta.174
Conclusion. There is increasing evidence both in vivo
and in vitro that arginine can improve endothelial
dysfunction and NO production in diabetes. Because of certain
methodological considerations, additional studies are warranted to
determine whether there are important regional differences in vascular
response to arginine and whether arginine will be efficacious at
various stages of the disease.
|
Effects of Arginine on NO in Human Diabetes
|
|---|
The possibility that NO deficits occur in clinical diabetes arises
from a report showing that arginine increases plasma nitrate/nitrite
levels in diabetic patients.200 Unfortunately,
the effects of arginine in control patients was not evaluated. In one
study, arginine increased nitrates and exhaled NO by nearly equivalent
degrees in control and IDDM patients.163 The
effects of arginine administration in vivo must be interpreted with
caution because arginine also increases insulin
secretion201 and the arginine stimulus might be
altered by disease.202 Insulin is reported to
cause dilation via stimulation of NO
production,203 204 205 although one recent
report shows that insulin can produce dilation in the presence of an
NOS inhibitor or can be blocked by adenosine
receptor antagonists or
KATP-sensitive channel
blockers.206
To circumvent this limitation, one study in patients with uncomplicated
IDDM revealed that arginine treatment augmented plasma cGMP levels and
L-citrulline levels to a similar extent in both control and
IDDM patients.178 This is consistent with
data showing that arginine supplementation improves
endothelium-dependent relaxation in
NIDDM.207 In contrast, high concentrations of
L-arginine had no effect in IDDM
patients.208 Arginine has also been shown in
NIDDM patients but not healthy controls to enhance insulin-mediated
dilation.209
Conclusion. While positive results indicate that
arginine may reverse NO deficits in humans, additional studies are
required to confirm or refute these initial findings.
|
Are Effects of Arginine NO/cGMP–Dependent?
|
|---|
It has been widely assumed that the
L-arginine–induced improvement in
endothelium-dependent relaxation in various other
models of vascular disease reflects enhanced NO/cGMP
production. The fallacy of this assumption remains to be
rigorously tested in diabetic models, since NO can elicit
cGMP-independent effects in some tissue. For example,
L-arginine augmented endothelium-dependent
relaxation in hypertensive arteries via an NO-independent pathway
involving cyclooxygenase-derived reactive
oxygen.210 Furthermore, there are studies that
show that arginine analogues which inhibit NOS also alter prostanoid
synthesis211 and
KATP-mediated
relaxation.212 Thus, it is possible that arginine
supplementation might improve relaxation in diabetic arteries via
alternative NO-independent pathways.
In diabetic rat aorta, the improvement in relaxation to acetylcholine
by L-arginine could be elicited despite coadministration of
either TEA, indomethacin, or
catalase.127 This suggests that enhanced
K+ channel activity, vasodilator
prostaglandins, and
H2O2 cannot account for the
beneficial effects of L-arginine. More recent studies using
glibenclamide suggest that L-arginine also does not improve
relaxation via enhanced KATP-mediated relaxation
(G.M.P., unpublished observations, 1997). It is not possible to
conclude that similar results will be obtained in other arteries and
species, since a comprehensive analysis has not been
forthcoming. We do note that the beneficial effects of
L-arginine in diabetic canine coronaries occurred despite
the fact that all animals were treated with aspirin to inhibit
cyclooxgenase activity.20
In diabetic canine coronary arteries and in diabetic rat aorta,
restoration of endothelial function by
L-arginine cannot be accounted for by altered prostacyclin
production or enhanced K+ channel
activity. Combined with the data showing augmented cGMP generation,
this suggests that the effects of arginine in diabetes are specific for
enhanced NO activity.
|
Effects of Arginine on Endothelial Dysfunction
Produced by Elevated Glucose Concentration In Vitro or Hyperglycemia
In Vivo
|
|---|
One study noted that flow reduction in normal isolated guinea pig
hearts resulted in vasoconstriction that was reversed by
L-arginine.213 In contrast, the acute
vasodilation produced by flow reduction in the presence of 44
mmol/L D-glucose was augmented by L-arginine.
This suggests that arginine may play an important role in
autoregulation under high glucose conditions.
In another study, incubation of normal rat mesenteric arteries with
elevated glucose concentrations in vitro caused impaired relaxation to
acetylcholine.79 Addition of 0.1 mmol/L
L-arginine abrogated the effect of 20 mmol/L glucose
and significantly reduced the detrimental effects of 45 mmol/L
glucose. Another study showed that glucose-induced decrease in
Na+,K+-ATPase of normal
rabbit aorta was also prevented by coincubation with
L-arginine.214
The direct effects of glucose have also been examined in hyperglycemic
clamp experiments in leg blood flow of healthy
subjects.76 The decreased blood flow observed
after 60 and 90 minutes of hyperglycemia was completely reversed by
L-arginine. The fact that neither D-arginine or
L-lysine was effective illustrates the selectivity and
specificity for the L-arginine action.
Conclusion. Arginine appears to provide protection from
alterations in vascular function caused by elevated glucose
concentration. The mechanism of this protection is still unknown.
|
Potential Actions of Chronic Arginine Treatment on Endothelial
Function
|
|---|
Top
Introduction
Impaired Endothelium-Dependent...